In this report, we analyzed the hazard risk from coastal flooding in Menlo Park. Menlo Park is a city at the eastern edge of San Mateo County. The population is about 35,000. A considerable number of people live in areas close to the coastal. They are at great risk in the event of rising sea level and the occasional return of flood. The report mainly focus on the estimation of vehicles average annual loss caused by flood from 2020 to 2050.

Part 1 flood maps

To get the hazard scenarios, we used data from OCOF and cropped to the extent of Menlo Park. We considered the permutations and combinations of 3 SLRs(Sea Level Rise) and 3RPs(Return Period), a total of 9 scenarios.

Here, we took SLR=50 and RP=100 as an example, and drew the flood depth map.

It can be seen that when Sea Level Rise is 50 and Return Period is 100, some buildings in Menlo Park would be affected by floods, especially those in the northeast part.

Next, we repeated this process and obtained a total of 9 maps. Please check this link to see the specific content of these maps: https://hhyj4495.shinyapps.io/dashboard_flood_menlo_park/

Part 2 Get the number of vehicles per building in Menlo Park

So far, we have collected the hazard data for Menlo Park. In order to get the number of vehicles, we did the following steps: First, using the latest ACS 5-yr data about vehicle ownership in Menlo Park, and produce an estimate of the total number of owned vehicles. Second, using the EMFAC data to estimate the growth of vehicles. We assumed that the % vehicle ownership rate doesn’t change over the next 30 years.

Then, we identified the census blocks within Menlo Park, and use OpenStreetMap data to retrieve all building footprints within these blocks. Assume that the exposure based on building footprints does not change over the study period. Then, we use 2020 Decennial census data to calculate the total population in each block.

EMFAC
Year Vehicle Category Fuel Type Population percentage
2020 LDA Gasoline 2698816 1.000000
2030 LDA Gasoline 2714691 1.005883
2040 LDA Gasoline 2818585 1.044379
2050 LDA Gasoline 2962226 1.097602
menlo park population in each block
GEOID20 pop
060816121011000 25
060816121011001 34
060816121011003 133
060816121011004 70
060816121011006 9
060816121011007 17

We assumed that buildings with type != NA are all residential buildings. This is because non-residential buildings are marked with their corresponding types in OpenStreetMap data set. Then we allocated 2020 vehicles from the whole CBG to each building, assuming population is distributed evenly across buildings in a block, and vehicles are distributed evenly across population.

menlo park blocks, vehicle per building
GEOID20 cbg bldg_count pop vehicle_count veh_per_person ppl_per_bldg veh_per_bldg
060816116001001 060816116001 55 82 774 0.6604096 1.490909 0.9846106
060816116001002 060816116001 41 75 774 0.6604096 1.829268 1.2080663
060816116001003 060816116001 40 81 774 0.6604096 2.025000 1.3373294
060816116001004 060816116001 57 104 774 0.6604096 1.824561 1.2049578
060816116001005 060816116001 40 101 774 0.6604096 2.525000 1.6675341
060816116001006 060816116001 20 55 774 0.6604096 2.750000 1.8161263

Base on this, we got the number of vehicles corresponding to each building in Menlo Park. We created a data frame to save these information. For each building, we recorded its osm_id, the number of vehicles of this building, and its location.

Next, we collected vulnerability data on the relationship between flood depth and vehicle damage from: https//planning.erdc.dren.mil/toolbox/library.cfm?Option=Listing&Type=EGM&Search=Policy&Sort=Default

We took the data of vehicle type = sedans to simplify the calculation.

vulnerability, flood depth and vehicle damage
depth perc_damage moe
0.0 0.000 0.0000
0.5 0.076 0.0242
1.0 0.280 0.0184
2.0 0.462 0.0151
3.0 0.622 0.0145
4.0 0.760 0.0157
5.0 0.876 0.0174
6.0 0.970 0.0192
7.0 1.000 0.0206
8.0 1.000 0.0206
9.0 1.000 0.0206
10.0 1.000 0.0206

The table above states that if the depth above ground is 3, then the percentage of damage would be 62.2%.

After obtaining all the above information, we got the average flood depth of each building under different SLR and RP conditions. Next, we took the flood depth of each building as input values, and obtained the corresponding percent damage from the table vulnerability, flood depth and vehicle damage by means of linear interpolation.

Assumed that all vehicles are parked on the ground floor (no basement, no underground parking), so the flood depth suffered by the building is equivalent to the flood depth suffered by the vehicles.

Next, we created an interactive plot to show the relationship between damaged vehicles and PR/SLR. In order to do that, a data frame is created, which has every combination of OpenStreetMap ID, sea level rise, and return period.

As is shown in the plot above, we can drag the pointer to set the SLR to 0, 25 or 50. It can be seen that when the SLR is 25m the average of percent damage is about 30%-40%, while when the SLR is 50, the average of percent damage exceeds 50%.

In the next step, we quantified these losses and estimated the average annualized loss in $ vehicle damages in Menlo Park from 2020 to 2050. To simplify the calculation, we made several assumptions: 1) The average cost of owning a car is $14,571 according to a U.S. News and World Report study. 2) Pickup trucks accounted for 20.57 percent of all vehicles in operation, according to analysis by Experian Automotive. The data can be found in Experian Automotive’s AutoCount Vehicles in Operation database. So we assume that 20.57% of the vehicles are immune to the hazard. 3) We assume that 25% of the vehicles are likely to be moved away from the hazard exposure with advanced warning.

Therefore: \[vehicle\ damage\ in\ USD = (1-percent\ move) * (1-percent\ immune) *cost\ per\ vehicle* percent\ damage \]

menlo park, $ vehicle damages per building
osm_id SLR damage
123925032 000 0.2258485
123925065 000 0.4893384
48232776 000 21.1813977
123925034 000 295.2190495
123925051 000 32.5361313
123925061 000 302.8075703

Here, we use RCP 4.5 occurrence rates of sea level rise in the Bay Area across years.

RCP4.5
SLR 2020 2030 2040
0 0.942 0.923 0.793
25 0.000 0.051 0.198
50 0.000 0.000 0.001
75 0.000 0.000 0.000
100 0.000 0.000 0.000
125 0.000 0.000 0.000
150 0.000 0.000 0.000
175 0.000 0.000 0.000
200 0.000 0.000 0.000
500 0.000 0.000 0.000

When predicting $ vehicle damages, we mainly considered two factors, one is the increase in the number of vehicles, and the other is the change in SLR. The change in the number of vehicles has been estimated by EMFAC, and the change of SLR could be estimated from the above table, which has the probability of each level of SLR. Based on this, we could get the annual average loss of vehicles of each building in 2020, 2030, 2040 and 2050.

In the above map, we took the difference between the AALs in 2050 and 2020. As is shown, the AAL in 2050 is significantly higher than the AAL in 2020. In the coastal area of Menlo Park, the AALs are larger in the northwest, where some buildings have AALs that have grown by close to $2,000. By contrast, the AALs in the southeast are relatively negligible, and almost all buildings have changes in AALs that are less than 100USD.

We summarized these buildings into the CBGs they belong to, and got the above map. CBG in the southwest of Menlo Park has the largest AAL, and its average annualized loss across 99 buildings is $35,000 from 2020 to 2050. This may be due to the high density of vehicles, or the high flood depth. In order to solve these problems and avoid potential damages, adding urban drainage facilities or multistorey car park might be necessary.

Part 3 Supplementary Analysis

As we found that table vulnerability, flood depth and vehicle damage contains information about the standard deviations, we performed a Monte Carlo simulation to propagate the uncertainty in the depth-damage relationship. For each building, we performed 1000 simulations.

Then, we took the average of perc_damage from 1000 simulations. In this way, the damage percentage of vehicles of each building is obtained.

perc_damage of vehicles, under Monte Carlo simulation
perc_damage avg_depth osm_id SLR RP
69636 0.0000000 0.0000000 123925032 000 001
69645 0.0000000 0.0000000 123925065 000 001
25570 0.0008310 0.0055747 48232776 000 020
696361 0.0000000 0.0000000 123925032 000 020
69637 0.0517056 0.3468819 123925034 000 020
69641 0.0000000 0.0000000 123925051 000 020
perc_damage of vehicles
perc_damage avg_depth osm_id SLR RP
69636 0.0000000 0.0000000 123925032 000 001
69645 0.0000000 0.0000000 123925065 000 001
25570 0.0008474 0.0055747 48232776 000 020
696361 0.0000000 0.0000000 123925032 000 020
69637 0.0527261 0.3468819 123925034 000 020
69641 0.0000000 0.0000000 123925051 000 020

There is a certain degree of change in perc_damage before and after Monte Carlo simulation, bt due to the small MOE, the change in perc_damage is not that sensitive to it. However, the resistance of the vehicle to flood depth is still a very important factor as it directly affects the AAL.

Also, there are other sources of uncertainty exist but outside the scope of our analysis.

Uncertainties - % vehicle damamge
factor1 Improvement of vehicle waterproofing system possitive
factor2 Upgrading of urban water supply and drainage system possitive
factor3 Relocation of housing in coastal areas possitive
factor4 # multistorey car park positive
factor5 Increased global warming leads to more dramatic sea level rise negative
factor6 Return period shorteded negative

We found that the AAL is not that sensitive to vulnerability, flood depth and vehicle damage, and AAL would not be affected much after 1000 simulations. It might because AAL is more affected by RP, SLR, as well as population growth, rising number of vehicles, rising vehicle prices and other factors.

The following two tables list the changes of the 6 buildings before and after Monte Carlo simulation.

Monte Carlo, aal_by_year_map
osm_id 2020 2030 2040 2050 change
123925032 181.5261 243.7674 423.2651 818.3913 636.8651
123925065 175.4775 234.4028 404.1044 777.2328 601.7553
48232776 822.2192 947.2753 1278.4944 1952.3367 1130.1175
123925034 1500.5869 1661.5731 2059.7681 2820.3242 1319.7373
123925051 992.6452 1128.5814 1482.2611 2188.9755 1196.3303
123925061 1248.2215 1390.3225 1746.7622 2439.7866 1191.5651
aal_by_year_map
osm_id 2020 2030 2040 2050 change
123925032 184.5968 247.1031 427.2173 823.4194 638.8226
123925065 178.4902 237.6897 408.0396 782.3248 603.8345
48232776 825.7069 950.9014 1282.3216 1956.2437 1130.5368
123925034 1503.6081 1664.5793 2062.5386 2822.3582 1318.7501
123925051 995.3785 1131.4171 1485.2357 2191.9687 1196.5902
123925061 1252.6926 1394.8350 1751.1198 2443.4532 1190.7606

We noticed that there are many households with only 1 vehicle or no vehicle. So we estimate the number of 0-vehicle and 1-vehicle households in Menlo Park that are exposed to some amount of flood risk.

The map below showed the distribution of such household, from 2020 to 2050.

The CBG in the northwest corner of menlo park has a maximum of 0 vehicle or no vehicle households. During flood period, although these households would not be burdened by economic losses caused by vehicle damage, they do not have the necessary conditions for escaping during flood period. For these households, escape alternatives, such as public transportation, should be provided.

At last, we performed an equity analysis on the flood-affected households. We consider households with an income < $45,000 to be more vulnerable because their disposable income, say, total income minus basic living expense, is low, so they are more likely to be burdened by flood damage.

The census block groups that will be affected by flood are: 060816117001,060816117002,060816117003,060816117004.

In the plot above, we compared the total population in Menlo Park with the population of CBGs that will be affected by flooding, an it can be seen that low income (<$45,000) households are more likely to be affected by flooding. There seems to be a disproportionate burden of flooding on low income households. Therefore, additional assistance may be required for these households, or it would be necessary to help them to move away from the hazard exposure to avoid damage.